Abstract
This study proposes a linear regression-based intelligent calibration approach to improve the accuracy of water turbidity sensor readings in an Arduino microcontroller-based embedded system. The primary objective of this research is to develop a mathematical model capable of converting analog values from the Analog-to-Digital Converter (ADC) into a numerical representation that reflects the actual water turbidity level in Nephelometric Turbidity Units (NTU). The calibration process was performed using a standard Hanna Turbidity Meter with water samples ranging from 0.56 NTU to 500 NTU. Measurement results demonstrated a strong linear relationship between the ADC value (495–686) and NTU, with an average system accuracy level above 90%. Comparison of sensor measurements with the standard instrument showed an error margin below 5%, confirming the reliability of the linear regression model in compensating for optical sensor nonlinearities.
